# [gmx-users] Berendsen Temperature Coupling and Heat

David van der Spoel spoel at xray.bmc.uu.se
Wed Jul 30 09:35:34 CEST 2008

```Daniel S. Han wrote:
> Hello,
> I am running a steered MD simulation in an NPT ensemble using Berendsen
> thermostats.  I expect that my system could heat up due to the external
> forces, and this heat would eventually be absorbed by the heat bath.
>
> I thought I could get an estimate of the heat released by looking at the
> Temperature (T) and lambda values (velocity scaling factors) available
> in the gromacs energy output files, but I wasn't sure I was calculating
> it correctly.
>
> Here's the algorithm I thought might work:
> at every time step, we have a Temperature (T) and a Lambda (L) taken
> from the edr output file.
>
> Where we can calculate the kinetic energy at any step as:
> Kinetic Energy (KE) = .5 * N_degrees_of_freedom * K_boltzman * T
> (which I'll call KE_output)
>
> If I understand correctly, KE_output is the result after scaling by
> Lambda^2.  So if there were no temperature scaling, then the kinetic
> energy would be equal to:
> KE_no_scaling = KE_output / (lambda*lambda)
>
> Thus the heat exchanged between the system and the bath is:
> Heat = KE_output - KE_no_scaling
> where if heat is negative, it means that heat was released by the system
> and absorbed by the bath.
>
> Using this algorithm I found some odd results, where I found that both
> my Steered MD simulations and a simulation of the same system without
> external forces tended to absorb heat.
> I can post more details of that later, but for now, if anyone has any
> corrections/comments to this line of thought, it would be much appreciated.
>

It seems your appraoch is right. You can just plot lambda using
g_energy. If it less than one on average you are cooling, if it is more
than one you are heating the system.

If you integrate lambda - 1 you can compute the net heat flux into/out
of the system.

> Danny.
>
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the www
> interface or send it to gmx-users-request at gromacs.org.